Sleep, 14(6):540--545 © 1991 American Sleep Disorders Association and Sleep Research Society
A New Method for Measuring Daytime Sleepiness: The Epworth Sleepiness Scale Murray W. Johns
Summary: The development and use of a new scale, the Epworth sleepiness scale (ESS), is described. This is a simple, self-administered questionnaire which is shown to provide a measurement of the subject's genera1level of daytime sleepiness. One hundred and eighty adults answered the ESS, including 30 normal men and women as controls and 150 patients with a range of sleep disorders. They rated the chances that they would doze off or fall asleep when in eight different situations commonly encountered in daily life. Total ESS scores significantly distinguished normal subjects from patients in various diagnostic groups including obstructive sleep apnea syndrome, narcolepsy and idiopathic hypersomnia. ESS scores were significantly correlated with sleep latency measured during the multiple sleep latency test and during overnight polysomnography. In patients with obstructive sleep apnea syndrome ESS scores were significantly correlated with the respiratory disturbance index and the minimum Sa0 2 recorded overnight. ESS scores of patients who simply snored did not differ from controls. Key Words: SleepinessQuestionnaire-Sleep propensity-Insomnia-Obstructive sleep apnea syndrome.
A large proportion of adult patients who present to sleep disorder centers have disorders associated with excessive daytime sleepiness. These include obstmctive sleep apnea syndrome (OSAS), periodic limb movement disorder (PLMD), narcolepsy, idiopathic hypersomnia and other miscellaneous disorders (I). The severity of their chronic daytime sleepiness is an important aspect of each patient's assessment. Thus, there is a great need for a simple standardized test for measuring a patient's general level of sleepiness, which is independent of short-term variations in sleepiness, with the time of day and from day to day. The multiple sleep latency test (MSLT) is widdy used and is generally believed to provide a valid measurement of sleepiness on the particular day of the test (2,3). It is based on the premise that the sleepier lthe subject, the quicker he will fall asleep when encouraged to do so while lying down in a non stimulating environment. The MSLT has a reasonably high test-retest reliability over periods of months in normal subjects (4). Assuming the same reliability holds tme for patients, the MSLT must be considered the standard method for measuring their chronic daytime sleepiness. However, the MSLT is very cumbersome, timeAccepted for publication July 1991. Address correspondence and reprint requests to Dr. Murray w. Johns, Sleep Disorders Unit, Epworth Hospital, Melbourne, Victoria 3121, Australia.
consuming and expensive to perform. It takes all day, both for the subject and the polysomnographer and is not easy to justify as a routine test for all patients. Other measures of sleepiness have been devised (5,6). In the maintenance of wakefulness test (MWT) the latency to sleep onset is measured with the subject sitting in a dimly lit, warm, quiet room, trying to stay awake rather than to fall asleep (5). However, all such tests share the disadvantage of the MSLT in being cumbersome and expensive. Similar criticisms can be levelled at tests of sleepiness based on pupillometry (7), or cerebral evoked potentials (8). Other assessments of sleepiness have involved prolonged psychomotor performance tests, the results of which are not related in any simple or consistent way to sleepiness in different subjects (9). By contrast, the Stanford sleepiness scale (SSS) is a quick and simple test (l0). It involves the subject's own reports of symptoms and feelings at a particular time. Visual analogue scales (V AS) of sleepiness/alertness have also been used in this context (11). However, these tests do not attempt to measure the general level of daytime sleepiness, as distinct from feelings of sleepiness at a particular time. Nor, it appears, is the subjective sleepiness that they measure the same as the objective sleepiness measured by the MSLT (3,7). Scores on the SSS or on a VAS of sleepiness are not significantly correlated with sleep latency in the MSLT,
540
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
Sleep Disorders Unit, Epworth Hospital, Melbourne, Victoria, Australia
A SCALE MEASURING DAYTIME SLEEPINESS TABLE 1. The Epworth sleepiness scale THE EPWORTH SLEEPINESS SCALE Name: Today's date: Your sex (male
= M; female = F):
Your age (years): _ _ __ _ _ _ _ _ _ _ _ __
Situation Sitting and reading Watching TV Sitting, inactive in a public place (e.g. a theater or a meeting) As a passenger in a car for an hour without a break
Chance of dozing
Lying down to rest in the afternoon when circumstances permit Sitting and talking to someone Sitting quietly after a lunch without alcohol In a car, while stopped for a few minutes in the traffic Thank you for your cooperation
even when measured at virtually the same time (12). These subjective reports may be related more to tiredness and fatigue than to sleep propensity, as manifested by the tendency to fall asleep. The present report describes the development and use of a new questionnaire, the Epworth sleepiness scale (ESS), designed to measure sleep propensity in a simple, standardized way. The scale covers the whole range of sleep propensities, from the highest to the lowest.
Development of the ESS The concept of the ESS was derived from observations about the nature and occurrence of daytime sleep and sleepiness. Some people who suffer from excessive daytime sleepiness keep themselves busy and choose not to lie down nor to sit and relax during the day, thereby purposely avoiding daytime sleep. Others who may be bored, with spare time or who are socially withdrawn but who may not be very sleepy, choose to lie down and sleep during the day. About 50% of ostensibly healthy medical students usually sleep during the day at least once in an average week (13). Among 17-22-year-old recruits entering the French army, 19% reported sleeping during the day, regularly or occasionally. But only 5% complained of daytime sleepi-
ness (14). Thus, knowing how frequently or for how long subjects usually sleep during the day will probably not provide a useful measurement of their sleepiness. By contrast, sleepy people often describe how they doze off inadvertently while engaged in activities that involve low levels of stimulation, relative immobility and relaxation, such as sitting and watching TV. Earlier questionnaire surveys have indicated which situations, commonly encountered in daily life, are the most soporific (15). A large survey among adults in New Mexico asked about their frequency of falling asleep in five situations (16). The authors derived a score from the three "most sleepy" questions, which referred to falling asleep while "inactive in a public place", "at work", and "in a moving vehicle as passenger or driver". MSLTs on 116 of these subjects showed a statistically significant correlation between their sleep latency (SL) and their answers to those three questions (r = -0.32, p < 0.001). The ESS is based on questions referring to eight such situations, some known to be very soporific; others less so. The questionnaire, which is self-administered, is reproduced in Table 1. Subjects are asked to rate on a scale of 0-3 how likely they would be to doze off or fall asleep in the eight situations, based on their usual way of life in recent times. A distinction is made between dozing off and simply feeling tired. If a subject has not been in some of the situations recently, he is asked, nonetheless, to estimate how each might affect him. The ESS tries to overcome the fact that people have different daily routines, some facilitating and others inhibiting daytime sleep. For example, the ESS does not ask how frequently the subject falls asleep while watching TV. That would depend on how frequently he watched TV as much as on his sleepiness. Instead, the subject rates the chances that he would doze off whenever he watches TV. One question asks how likely the subject would be to doze off while lying down to rest in the afternoon when circumstances permit. It was felt that normal people probably would, and sleepy people certainly would tend to doze off in that situation. Never to do so would indicate an unusually low level of sleepiness, as described by some insomniacs. Some other situations were included in the questionnaire because it was believed that only the most sleepy people.would doze in them - while sitting and talking to someone, and in a car while stopped for a few minutes in traffic. These suppositions proved correct. The numbers selected for the eight situations in the ESS were added together to give a score for each subject, between 0 and 24. These ESS scores proved capable of distinguishing individuals and diagnostic groups over the whole range of daytime sleepiness. Sleep. Vol. 14, No.6, 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
How likely are you to doze off or fall asleep in the following situations, in contrast to feeling just tired? This refers to your usual way oflife in recent times. Even if you have not done some of these things recently try to work out how they would have affected you. Use the following scale to choose the most appropriate number for each situation: o = would never doze I = slight chance of dozing 2 = moderate change of dozing 3 = high chance of dozing
541
542
M. W.JOHNS TABLE 2. Subjects/diagnoses
The groups of experimental subjects, their ages and ESS scores Total number of subjects (M/F)
Normal controls Primary snoring OSAS . Narcolepsy Idiopathic hypersomnia Insomnia PLMD
36.4 45.7 48.4 46.6 41.4 40.3 52.5
± ± ± ± ± ±
9.9 10.7 10.7 12.0 14.0 14.6 + 10.3
ESS scores (mean + SD)
Range
5.9 ± 2.2 6.5 ± 3.0 11.7 ± 4.6 17.5 ± 3.5 17.9±3.1 2.2 ± 2.0 9.2 + 4.0
2-10 0-11 4-23 13-23 12-24 0-6 2 16
had excessive daytime sleepiness due to OSAS. The ESS scores for the 27 patients who had MSLTs ranged Subjects from 11 to 24. All patients with primary snoring had presented iniA total of 180 adult subjects completed the qUI~s tially because of the intensity and persistence of their tion~aire. There were 30 controls who were mainly snoring, on most nights at least. Many had been obhospItal employees, working during the day, who gave served .at home to pause in their breathing at night, a history of normal sleep habits without snoring. There su~estmg that they may have had sleep apnea, but were 150 patients with various sleep disorders, whose ages, sex and diagnostic categories are shown in Table thIS was found not to be of clinical significance by 2. Every new patient who presented to the Epworth polysomnography. The respiratory disturbance index Sleep Disorders Unit answered the ESS at their first (RDI) was calculated as the number of apneas and consultation. After investigation, all patients with the hypopneas causing a drop of > 3% in the arterial oxdiagnoses listed in Table 2 were included in the study ygen saturation per hour of sleep. The RDI for primary until there were 150. The ages of patients ranged from snorers was ::55. The 55 patients with OSAS were di18 to 78 years. The mean age within diagnostic groups vided into three subcategories according to their RDI, varied from 36 to 52 years. Men greatly outnumbered regardless of their complaints about daytime sleepiness women in the snoring, OSAS and PLMD groups. The or insomnia (Table 3). The RDI for mild OSAS was sexes were about equal in the other groups, apart from within the range> 5-15; for moderate OSAS the range was > 15-30, and for severe OSAS it was > 30. the insomniacs where women outnumbered men. A diagnosis of PLMD was made only if there were A total of 138 patients had overnight polysomnogat least 90 separate movements in one or both legs per raphy, but another 12 who were clearly suffering from night. The mean periodic movement index for these either chronic psychophysiological or idiopathic linsubjects, calculated as the number of movement events somnia did not. The latter diagnoses were made on the per hour of sleep, was 43.6 ± 30.4 (SD). Patients who basis of each patient's history, using the criteria set out had both PLMD and OSAS were excluded from this in the International Classification of Sleep Disordl~rs study. However, 9 of the 18 subjects with PLMD snored (1). Other insomniacs, with mood disorders or drug during polysomnography without having OSAS. effects, were excluded. Twenty-seven patients had MSLTs after overnight polysomnography. They had four naps, at 1000, 1200, Statistical methods 1400 and 1600 hours. Sleep latency was measured from The ESS scores of male and female control subjects the time lights were switched off until the onset of stage were compared by a Student's t test. Differences in ESS 1 sleep of at least 1 minute duration, or the onset of scores between the diagnostic groups were tested by either stage 2 or rapid eye movement (REM) slec!p. one-way ANOYA and then by posthoc ScheWe tests. The early onset of REM sleep was indicated by the A separate ANOYA and posthoc ScheWe tests were occurrence of REM sleep within 20 minutes of sle:ep onset. Of the 27 patients, 11 had narcolepsy diagnosed from the patient's history, particularly of cataplexy, TABLE 3. ESS scores in mild, moderate and severe OSAS associated with an SL of less than 10 minutes and the Total early onset of REM sleep in two or more naps (10 number of patients) or in one nap (1 patient with cataplexy). FourMean subjects ESS scores (M/F) (mean ± SD) Range RDI ± SD teen of the 27 patients had idiopathic hypersomnia, diagnosed from their excessive daytime sleepiness in MildOSAS 8.8 ± 2.3 22 (21/1) 9.S ± 3.3 4-16 21.1 ± 4.0 20 (20/0) 11.5 ± 4.2 5-20 the absence of either cataplexy or the early onset of Moderate OSAS 49.5 + 9.6 13 (12/1) 16.0 + 4.4 8 23 REM sleep in the MSLT. The remaining two patients Severe OSAS Sleep, Vol. 14. No.6, 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
METHODS
30 (14/16) 32 (29/3) 55 (53/2) 13 (8/5) 14 (8/6) 18(6/12) 18 (16/2)
Age in years (mean + SD)
A SCALE MEASURING DAYTIME SLEEPINESS
RESULTS The mean ESS score for control subjects was 5.9 ± 2.2 (SD) and their modal score was 6. There was no significant difference in the scores between male and female controls (males = 5.64 ± 2.56; females = 6.06 ± 1.84, t = 0.520, p = 0.607). Consequently, no distinction was made between the sexes in other groups. Patients suffering from disorders known to be associated with excessive daytime sleepiness reported the likelihood of dozing under circumstances that were not conducive to sleep in normal subjects. For example, 96% of the patients with either narcolepsy or idiopathic hypersomnia reported some chance, and often a high chance, of dozing while sitting and talking to someone, or in a car while stopped for a few minutes in the traffic. Only 6% of controls reported a slight chance of doing so. Patients with persistent psychophysiological or idiopathic insomnia reported either a complete inability or only a slight chance of dozing while lying down to rest in the afternoon when circumstances permitted. By contrast, 94% of controls reported some likelihood of dozing then. One-way ANOVA demonstrated significant differences in ESS scores between the seven diagnostic groups in Table 2 (F= 50.00; df= 6,173; p < 0.0001). Posthoc tests between paired groups showed that the ESS scores for primary snorers did not differ from controls (p = 0.998). Scores for OSAS, narcolepsy and idiopathic hypersomnia were significantly higher than for controls (p < 0.001) or primary snorers (p < 0.001). The insomniacs had significantly lower scores (p < 0.01) than all groups other than controls, for which the difference did not quite reach statistical significance (p = 0.063). The ESS scores of patients with PLMD did not differ significantly from controls (p = 0.149). A separate one-way ANOVA for the ESS scores of primary snorers and the three subcategories of OSAS showed significant differences between these groups (F = 23.11; df = 3,82; p < 0.001). Posthoc tests then showed that ESS scores for each level of OSAS were
significantly higher than for primary snorers (p = 0.035 for mild OSAS; p < 0.001 for moderate and severe OSAS). Scores for severe OSAS were higher than for moderate OSAS (p < 0.001), but the difference between mild and moderate OSAS did not reach statistical significance (p = 0.085). Considering a1155 patients with OSAS together, there was a significant correlation, on the one hand, between ESS scores and RDI (r = 0.550, p < 0.001) and on the other hand, between ESS scores and the minimum Sa0 2 recorded during apneas overnight (r = -0.457, p < 0.001). The RDI and the minimum overnight Sa02 during apneas were also significantly correlated (r = -0.687, p < 0.001). The linear regression equations for these three relationships, in the form Y = a + bx, were as follows: (RDI) = -0.674 + 2.006(ESS score) (minimum Sa02%) = 86.47 - 1.055(ESS score) (minimum Sa02%) = 84.15 - 0.440(RDI) Among the 138 patients who had overnight polysomnography there was a significant correlation between ESS score and (In) sleep latency at night (r = -0.379, n = 138, p < 0.001). In the smaller group of patients who had MSLTs, the correlation between In (SL) during the day and ESS score was also statistically significant (r = -0.514, n = 27, p < 0.01. The linear regression equation for this relationship was In (SL) = 3.353 - 0.091(ESS score). Individual ESS scores of 16 or more, indicating a high level of daytime sleepiness, were found only in patients with narcolepsy, idiopathic hypersomnia or OSAS of at least moderate severity (i.e. RDI > 15). All patients with either narcolepsy or idiopathic hypersomnia had higher ESS scores than the controls (i.e. ESS > 10) as did 12 of 13 patients with severe OSAS. The remaining patient in the latter category had an ESS score of 8 and was clinically not much affected by his sleep apnea. Within the group of patients with PLMD, the periodic movement index, which ranged from 16 to 122 movements per hours of sleep, was not significantly correlated with ESS scores (r = 0.049, n = 18, p > 0.1).
DISCUSSION These results provide evidence that a questionnairebased scale as brief and as simple as the ESS can give valid measurements of sleep propensity in adults. ESS scores significantly distinguished groups of patients who are known from other investigations to have differences in their levels of sleepiness, as measured by the MSLT (2,18). ESS scores were significantly correlated Sleep. Vol. 14. No.6. 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
used to test the differences in ESS scores between primary snorers and the three categories of OSAS. The ScheWe test is conservative and is suitable for groups with unequal numbers of subjects (17). The distribution of sleep latencies ,measured in minutes, was highly skewed positively and was normalized by log., transformation. The relationships between pairs of continuous variables, such as RDI and sleep latency during overnight polysomnography, were tested by Pearson correlation coefficients and linear regression. Statistical significance was accepted at p < 0.05 in two-tailed tests.
543
544
M. W. JOHNS
with sleep latency measured during the day with MSLTs and at night with polysomnography. This is despite any effect of the first night in the laboratory. Others have found a significant positive correlation between the SL at night and during the day in the same subject (19).
ESS scores greater than 16, indicative of a high level of daytime sleepiness, were encountered only in patients with moderate or severe OSAS (RDI > 15), narcolepsy or idiopathic hypersomnia. These disorders are known to be associated with excessive daytime sleepiness as measured by the MSLT (2,18). Newrtheless, high ESS scores, by themselves, are not diagnostic of a particular sleep disorder, any more than is an SL of 5 minutes in an MSLT. ESS scores were correlated with both the RDI and the minimum Sa02 recorded during polysomnography Acknowledgement: Irene Lehel assisted with the adminin patients with OSAS of differing severity. In the past, these measures of the severity of OSAS have been found istration of questionnaires to the control subjects. to be related to the SL in MSLTs in some, but not in 1. American Sleep Disorders Association. The international clasall investigations (18,20). The finding that ESS scores sification of sleep disorders. Rochester, MN, 1990. can distinguish patients who simply snore from those 2. Richardson G, Carskadon M, Flagg W, Van Den Hoed J, Dement W, Mitler M. Excessive daytime sleepiness in man: mulwith even mild OSAS is evidence for the sensitivity of tiple sleep latency measurements in narcoleptic vs. control subthe ESS. The questionnaire should be useful in elucijects. Electroencephalogr Clin NeurophysioI1978;45:621-7. dating the epidemiology of snoring and OSAS, and any 3. Carskadon MA, Dement WC. The multiple sleep latency test: what does it measure? Sleep 1985;5:S67-72. associated cardiovascular or cerebrovascular risks. 4. Zwyghuizen-Doorenbos A, Roehrs T, Schaeffer M, Roth T. TestPrevious investigations of this kind have tended to blur retest reliability of the MSLT. Sleep 1988;11:562-5. the distinction between primary snoring and OSAS 5. Mitler M, Gujavarty KS, Browman CPo Maintenance of wakefulness test: a polysomnographic technique for evaluating treat(21). ment in patients with excessive somnolence. Electroencephalogr In the patients with PLMD, the finding of an almost Clin Neurophysiol1982; 153:658-61. 6. Erman MK, Beckham B, Gardner DA, Roffwarg HP. The modzero correlation between their periodic movement inified assessment of sleepiness test (MAST). Sleep Res 1987; 16: dex and ESS scores suggests that whatever level of 550. daytime sleepiness is associated with PLMD, it is not 7. Pressman MR, Fry JM. Relationship of autonomic nervous system activity to daytime sleepiness and prior sleep. Sleep 1989; 12: related simply to the frequency oflimb movements. It 239-45. may be more closely related to the frequency of those 8. Broughton R, Aguirre M, Dunham W. A comparison of multiple movements producing arousal rather than those that and single sleep latency and cerebral evoked potential (P300) measures in the assessment of excessive daytime sleepiness in do not. This distinction was not made here and further narcolepsy-{:ataplexy. Sleep 1988;11:537-45. investigation is required to clarify this relationship. 9. Johnson LC, Spinweber CL. Gomez SA, Matteson LT. Daytime The low ESS scores of patients with idiopathic or sleepiness, performance, mood, nocturnal sleep: the effect of psychophysiological insomnia are consistent with evbenzodiazepine and caffeine on their relationship. Sleep 1990; 13: 121-35. idence that such patients have a low sleep propensity, 10. Hoddes E, Zarcone V, Smythe H, Phillips R, Dement WC. even when they are able to relax (22). It must not be Quantification of sleepiness: a new approach. Psychophysiology assumed, however, that this is necessarily so for other 1973;10:431-6. 11. Herbert M, Johns MW, Dor€: C. Factor analysis of analogue kinds of insomnia, such as with mood disorders. scales measuring subjective feelings before and after sleep. Br J The relatively wide range ofESS scores in the control Med PsychoI1976;49:373-9. subjects [2-10] is consistent with evidence that some 12. Cook Y, Schmitt F, Berry D, Gilmore R, Phillips B, Lamb D. The effects of nocturnal sleep, sleep disordered breathing and healthy adults, without recognizable sleep disordt~rs, periodic movements of sleep on the objective and subjective remain sleepier than others during the day (23). Such assessment of daytime somnolence in healthy aged adults. Sleep differences persist in MSLTs, even after extending the Res 1988;17:95. hours of nocturnal sleep to overcome possible sleep 13. Johns MW, Gay TJA, Goodyear MDE, Masterton JP. Sleep habits of healthy young adults: use of a sleep questionnaire. Br deprivation (24). The sleep propensity of a subject on J Prev Soc Med 1971;25:236-41. a particular day would be influenced by the quality and 14. Billiard M, Alperovitch A, Perot C, Jammes A. Excessive daytime somnolence in young men: prevalence and contributing duration of prior sleep or of sleep deprivation, the time factors. Sleep 1987; 10:297-305. of day, the presence of various sleep disorders, drug 15. Roth T, Roehrs T, Carskadon M, Dement W. Daytime sleepiness and alertness. In: Kryger MH, Roth T, Dement WC, eds. effects, the level of interest and motivation induced by Sleep, Vol. 14, No.6, 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
the situation at hand, as well as longer-term physiological differences. The ESS does not distinguish the nature oflong-term physiological or pathological processes that produce a particular level of sleep propensity. Other investigations, including overnight polysomnography, are required for that. The ESS assumes that subjects can remember whether or not and under what circumstances they have dozed off during the day as part of their "usual way of life in recent times". The present results suggest that most patients can give meaningful self reports about this aspect of their behavior and that their ESS scores provide a measurement of their general level of daytime sleepiness, from low to very high levels. This has not been achieved previously by any other published questionnaire.
A SCALE MEASURING DAYTIME SLEEPINESS
16. 17. 18. 19. 20.
Principles and practice of sleep medicine. Philadelphia: Saunders, 1989:14-23. Schmidt-Nowara WW, Wiggins CL, Walsh JK, Bauer C. Prevalence of sleepiness in an adult population. Sleep Res 1989;18: 302. Munro BH, Yisintainer MA, Page EB. Statistical methods for health care research. Philadelphia: J. B. Lippincott Co., 1986. Roehrs T, Zorick F, Wittig R, Conway W, Roth T. Predictors of objective level of daytime sleepiness in patients with sleeprelated breathing disorders. Chest 1989;95:1202-6. Kaplan J, Fredrickson PA, Renaux SR. Nighttime sleep latency as an indicator of daytime sleepiness. Sleep Res 1990; 19:241. Guilleminault C, Partinen M, Quera-Salva MA, Hayes B, De-
21. 22. 23. 24.
545
ment WC, Nino-Murcia G. Determinants of daytime sleepiness in obstructive sleep apnea. Chest 1988;94:32-7. Waller PC, Bhopal RS. Is snoring a cause of vascular disease: an epidemiological review. Lancet 1989;1:143-6. Stepanski E, Zorick F, Roehrs T, Young D, Roth R. Daytime alertness in patients with chronic insomnia compared with asymptomatic control subjects. Sleep 1988; II :54-60. Lavie P, Segal S. Twenty-four-hour structure of sleepiness in morning and evening persons investigated by ultrashort sleepwake cycle. Sleep 1989;12:522-8. Roehrs T, Timms Y, Zwyghuizen-Doorenbos A, Roth T. Sleep extension in sleepy and alert normals. Sleep 1989;12:449-57. Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
Sleep, Vol. 14, No.6, 1991
A New Method for Measuring Daytime Sleepiness: The Epworth Sleepiness Scale Murray W. Johns
Summary: The development and use of a new scale, the Epworth sleepiness scale (ESS), is described. This is a simple, self-administered questionnaire which is shown to provide a measurement of the subject's genera1level of daytime sleepiness. One hundred and eighty adults answered the ESS, including 30 normal men and women as controls and 150 patients with a range of sleep disorders. They rated the chances that they would doze off or fall asleep when in eight different situations commonly encountered in daily life. Total ESS scores significantly distinguished normal subjects from patients in various diagnostic groups including obstructive sleep apnea syndrome, narcolepsy and idiopathic hypersomnia. ESS scores were significantly correlated with sleep latency measured during the multiple sleep latency test and during overnight polysomnography. In patients with obstructive sleep apnea syndrome ESS scores were significantly correlated with the respiratory disturbance index and the minimum Sa0 2 recorded overnight. ESS scores of patients who simply snored did not differ from controls. Key Words: SleepinessQuestionnaire-Sleep propensity-Insomnia-Obstructive sleep apnea syndrome.
A large proportion of adult patients who present to sleep disorder centers have disorders associated with excessive daytime sleepiness. These include obstmctive sleep apnea syndrome (OSAS), periodic limb movement disorder (PLMD), narcolepsy, idiopathic hypersomnia and other miscellaneous disorders (I). The severity of their chronic daytime sleepiness is an important aspect of each patient's assessment. Thus, there is a great need for a simple standardized test for measuring a patient's general level of sleepiness, which is independent of short-term variations in sleepiness, with the time of day and from day to day. The multiple sleep latency test (MSLT) is widdy used and is generally believed to provide a valid measurement of sleepiness on the particular day of the test (2,3). It is based on the premise that the sleepier lthe subject, the quicker he will fall asleep when encouraged to do so while lying down in a non stimulating environment. The MSLT has a reasonably high test-retest reliability over periods of months in normal subjects (4). Assuming the same reliability holds tme for patients, the MSLT must be considered the standard method for measuring their chronic daytime sleepiness. However, the MSLT is very cumbersome, timeAccepted for publication July 1991. Address correspondence and reprint requests to Dr. Murray w. Johns, Sleep Disorders Unit, Epworth Hospital, Melbourne, Victoria 3121, Australia.
consuming and expensive to perform. It takes all day, both for the subject and the polysomnographer and is not easy to justify as a routine test for all patients. Other measures of sleepiness have been devised (5,6). In the maintenance of wakefulness test (MWT) the latency to sleep onset is measured with the subject sitting in a dimly lit, warm, quiet room, trying to stay awake rather than to fall asleep (5). However, all such tests share the disadvantage of the MSLT in being cumbersome and expensive. Similar criticisms can be levelled at tests of sleepiness based on pupillometry (7), or cerebral evoked potentials (8). Other assessments of sleepiness have involved prolonged psychomotor performance tests, the results of which are not related in any simple or consistent way to sleepiness in different subjects (9). By contrast, the Stanford sleepiness scale (SSS) is a quick and simple test (l0). It involves the subject's own reports of symptoms and feelings at a particular time. Visual analogue scales (V AS) of sleepiness/alertness have also been used in this context (11). However, these tests do not attempt to measure the general level of daytime sleepiness, as distinct from feelings of sleepiness at a particular time. Nor, it appears, is the subjective sleepiness that they measure the same as the objective sleepiness measured by the MSLT (3,7). Scores on the SSS or on a VAS of sleepiness are not significantly correlated with sleep latency in the MSLT,
540
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
Sleep Disorders Unit, Epworth Hospital, Melbourne, Victoria, Australia
A SCALE MEASURING DAYTIME SLEEPINESS TABLE 1. The Epworth sleepiness scale THE EPWORTH SLEEPINESS SCALE Name: Today's date: Your sex (male
= M; female = F):
Your age (years): _ _ __ _ _ _ _ _ _ _ _ __
Situation Sitting and reading Watching TV Sitting, inactive in a public place (e.g. a theater or a meeting) As a passenger in a car for an hour without a break
Chance of dozing
Lying down to rest in the afternoon when circumstances permit Sitting and talking to someone Sitting quietly after a lunch without alcohol In a car, while stopped for a few minutes in the traffic Thank you for your cooperation
even when measured at virtually the same time (12). These subjective reports may be related more to tiredness and fatigue than to sleep propensity, as manifested by the tendency to fall asleep. The present report describes the development and use of a new questionnaire, the Epworth sleepiness scale (ESS), designed to measure sleep propensity in a simple, standardized way. The scale covers the whole range of sleep propensities, from the highest to the lowest.
Development of the ESS The concept of the ESS was derived from observations about the nature and occurrence of daytime sleep and sleepiness. Some people who suffer from excessive daytime sleepiness keep themselves busy and choose not to lie down nor to sit and relax during the day, thereby purposely avoiding daytime sleep. Others who may be bored, with spare time or who are socially withdrawn but who may not be very sleepy, choose to lie down and sleep during the day. About 50% of ostensibly healthy medical students usually sleep during the day at least once in an average week (13). Among 17-22-year-old recruits entering the French army, 19% reported sleeping during the day, regularly or occasionally. But only 5% complained of daytime sleepi-
ness (14). Thus, knowing how frequently or for how long subjects usually sleep during the day will probably not provide a useful measurement of their sleepiness. By contrast, sleepy people often describe how they doze off inadvertently while engaged in activities that involve low levels of stimulation, relative immobility and relaxation, such as sitting and watching TV. Earlier questionnaire surveys have indicated which situations, commonly encountered in daily life, are the most soporific (15). A large survey among adults in New Mexico asked about their frequency of falling asleep in five situations (16). The authors derived a score from the three "most sleepy" questions, which referred to falling asleep while "inactive in a public place", "at work", and "in a moving vehicle as passenger or driver". MSLTs on 116 of these subjects showed a statistically significant correlation between their sleep latency (SL) and their answers to those three questions (r = -0.32, p < 0.001). The ESS is based on questions referring to eight such situations, some known to be very soporific; others less so. The questionnaire, which is self-administered, is reproduced in Table 1. Subjects are asked to rate on a scale of 0-3 how likely they would be to doze off or fall asleep in the eight situations, based on their usual way of life in recent times. A distinction is made between dozing off and simply feeling tired. If a subject has not been in some of the situations recently, he is asked, nonetheless, to estimate how each might affect him. The ESS tries to overcome the fact that people have different daily routines, some facilitating and others inhibiting daytime sleep. For example, the ESS does not ask how frequently the subject falls asleep while watching TV. That would depend on how frequently he watched TV as much as on his sleepiness. Instead, the subject rates the chances that he would doze off whenever he watches TV. One question asks how likely the subject would be to doze off while lying down to rest in the afternoon when circumstances permit. It was felt that normal people probably would, and sleepy people certainly would tend to doze off in that situation. Never to do so would indicate an unusually low level of sleepiness, as described by some insomniacs. Some other situations were included in the questionnaire because it was believed that only the most sleepy people.would doze in them - while sitting and talking to someone, and in a car while stopped for a few minutes in traffic. These suppositions proved correct. The numbers selected for the eight situations in the ESS were added together to give a score for each subject, between 0 and 24. These ESS scores proved capable of distinguishing individuals and diagnostic groups over the whole range of daytime sleepiness. Sleep. Vol. 14, No.6, 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
How likely are you to doze off or fall asleep in the following situations, in contrast to feeling just tired? This refers to your usual way oflife in recent times. Even if you have not done some of these things recently try to work out how they would have affected you. Use the following scale to choose the most appropriate number for each situation: o = would never doze I = slight chance of dozing 2 = moderate change of dozing 3 = high chance of dozing
541
542
M. W.JOHNS TABLE 2. Subjects/diagnoses
The groups of experimental subjects, their ages and ESS scores Total number of subjects (M/F)
Normal controls Primary snoring OSAS . Narcolepsy Idiopathic hypersomnia Insomnia PLMD
36.4 45.7 48.4 46.6 41.4 40.3 52.5
± ± ± ± ± ±
9.9 10.7 10.7 12.0 14.0 14.6 + 10.3
ESS scores (mean + SD)
Range
5.9 ± 2.2 6.5 ± 3.0 11.7 ± 4.6 17.5 ± 3.5 17.9±3.1 2.2 ± 2.0 9.2 + 4.0
2-10 0-11 4-23 13-23 12-24 0-6 2 16
had excessive daytime sleepiness due to OSAS. The ESS scores for the 27 patients who had MSLTs ranged Subjects from 11 to 24. All patients with primary snoring had presented iniA total of 180 adult subjects completed the qUI~s tially because of the intensity and persistence of their tion~aire. There were 30 controls who were mainly snoring, on most nights at least. Many had been obhospItal employees, working during the day, who gave served .at home to pause in their breathing at night, a history of normal sleep habits without snoring. There su~estmg that they may have had sleep apnea, but were 150 patients with various sleep disorders, whose ages, sex and diagnostic categories are shown in Table thIS was found not to be of clinical significance by 2. Every new patient who presented to the Epworth polysomnography. The respiratory disturbance index Sleep Disorders Unit answered the ESS at their first (RDI) was calculated as the number of apneas and consultation. After investigation, all patients with the hypopneas causing a drop of > 3% in the arterial oxdiagnoses listed in Table 2 were included in the study ygen saturation per hour of sleep. The RDI for primary until there were 150. The ages of patients ranged from snorers was ::55. The 55 patients with OSAS were di18 to 78 years. The mean age within diagnostic groups vided into three subcategories according to their RDI, varied from 36 to 52 years. Men greatly outnumbered regardless of their complaints about daytime sleepiness women in the snoring, OSAS and PLMD groups. The or insomnia (Table 3). The RDI for mild OSAS was sexes were about equal in the other groups, apart from within the range> 5-15; for moderate OSAS the range was > 15-30, and for severe OSAS it was > 30. the insomniacs where women outnumbered men. A diagnosis of PLMD was made only if there were A total of 138 patients had overnight polysomnogat least 90 separate movements in one or both legs per raphy, but another 12 who were clearly suffering from night. The mean periodic movement index for these either chronic psychophysiological or idiopathic linsubjects, calculated as the number of movement events somnia did not. The latter diagnoses were made on the per hour of sleep, was 43.6 ± 30.4 (SD). Patients who basis of each patient's history, using the criteria set out had both PLMD and OSAS were excluded from this in the International Classification of Sleep Disordl~rs study. However, 9 of the 18 subjects with PLMD snored (1). Other insomniacs, with mood disorders or drug during polysomnography without having OSAS. effects, were excluded. Twenty-seven patients had MSLTs after overnight polysomnography. They had four naps, at 1000, 1200, Statistical methods 1400 and 1600 hours. Sleep latency was measured from The ESS scores of male and female control subjects the time lights were switched off until the onset of stage were compared by a Student's t test. Differences in ESS 1 sleep of at least 1 minute duration, or the onset of scores between the diagnostic groups were tested by either stage 2 or rapid eye movement (REM) slec!p. one-way ANOYA and then by posthoc ScheWe tests. The early onset of REM sleep was indicated by the A separate ANOYA and posthoc ScheWe tests were occurrence of REM sleep within 20 minutes of sle:ep onset. Of the 27 patients, 11 had narcolepsy diagnosed from the patient's history, particularly of cataplexy, TABLE 3. ESS scores in mild, moderate and severe OSAS associated with an SL of less than 10 minutes and the Total early onset of REM sleep in two or more naps (10 number of patients) or in one nap (1 patient with cataplexy). FourMean subjects ESS scores (M/F) (mean ± SD) Range RDI ± SD teen of the 27 patients had idiopathic hypersomnia, diagnosed from their excessive daytime sleepiness in MildOSAS 8.8 ± 2.3 22 (21/1) 9.S ± 3.3 4-16 21.1 ± 4.0 20 (20/0) 11.5 ± 4.2 5-20 the absence of either cataplexy or the early onset of Moderate OSAS 49.5 + 9.6 13 (12/1) 16.0 + 4.4 8 23 REM sleep in the MSLT. The remaining two patients Severe OSAS Sleep, Vol. 14. No.6, 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
METHODS
30 (14/16) 32 (29/3) 55 (53/2) 13 (8/5) 14 (8/6) 18(6/12) 18 (16/2)
Age in years (mean + SD)
A SCALE MEASURING DAYTIME SLEEPINESS
RESULTS The mean ESS score for control subjects was 5.9 ± 2.2 (SD) and their modal score was 6. There was no significant difference in the scores between male and female controls (males = 5.64 ± 2.56; females = 6.06 ± 1.84, t = 0.520, p = 0.607). Consequently, no distinction was made between the sexes in other groups. Patients suffering from disorders known to be associated with excessive daytime sleepiness reported the likelihood of dozing under circumstances that were not conducive to sleep in normal subjects. For example, 96% of the patients with either narcolepsy or idiopathic hypersomnia reported some chance, and often a high chance, of dozing while sitting and talking to someone, or in a car while stopped for a few minutes in the traffic. Only 6% of controls reported a slight chance of doing so. Patients with persistent psychophysiological or idiopathic insomnia reported either a complete inability or only a slight chance of dozing while lying down to rest in the afternoon when circumstances permitted. By contrast, 94% of controls reported some likelihood of dozing then. One-way ANOVA demonstrated significant differences in ESS scores between the seven diagnostic groups in Table 2 (F= 50.00; df= 6,173; p < 0.0001). Posthoc tests between paired groups showed that the ESS scores for primary snorers did not differ from controls (p = 0.998). Scores for OSAS, narcolepsy and idiopathic hypersomnia were significantly higher than for controls (p < 0.001) or primary snorers (p < 0.001). The insomniacs had significantly lower scores (p < 0.01) than all groups other than controls, for which the difference did not quite reach statistical significance (p = 0.063). The ESS scores of patients with PLMD did not differ significantly from controls (p = 0.149). A separate one-way ANOVA for the ESS scores of primary snorers and the three subcategories of OSAS showed significant differences between these groups (F = 23.11; df = 3,82; p < 0.001). Posthoc tests then showed that ESS scores for each level of OSAS were
significantly higher than for primary snorers (p = 0.035 for mild OSAS; p < 0.001 for moderate and severe OSAS). Scores for severe OSAS were higher than for moderate OSAS (p < 0.001), but the difference between mild and moderate OSAS did not reach statistical significance (p = 0.085). Considering a1155 patients with OSAS together, there was a significant correlation, on the one hand, between ESS scores and RDI (r = 0.550, p < 0.001) and on the other hand, between ESS scores and the minimum Sa0 2 recorded during apneas overnight (r = -0.457, p < 0.001). The RDI and the minimum overnight Sa02 during apneas were also significantly correlated (r = -0.687, p < 0.001). The linear regression equations for these three relationships, in the form Y = a + bx, were as follows: (RDI) = -0.674 + 2.006(ESS score) (minimum Sa02%) = 86.47 - 1.055(ESS score) (minimum Sa02%) = 84.15 - 0.440(RDI) Among the 138 patients who had overnight polysomnography there was a significant correlation between ESS score and (In) sleep latency at night (r = -0.379, n = 138, p < 0.001). In the smaller group of patients who had MSLTs, the correlation between In (SL) during the day and ESS score was also statistically significant (r = -0.514, n = 27, p < 0.01. The linear regression equation for this relationship was In (SL) = 3.353 - 0.091(ESS score). Individual ESS scores of 16 or more, indicating a high level of daytime sleepiness, were found only in patients with narcolepsy, idiopathic hypersomnia or OSAS of at least moderate severity (i.e. RDI > 15). All patients with either narcolepsy or idiopathic hypersomnia had higher ESS scores than the controls (i.e. ESS > 10) as did 12 of 13 patients with severe OSAS. The remaining patient in the latter category had an ESS score of 8 and was clinically not much affected by his sleep apnea. Within the group of patients with PLMD, the periodic movement index, which ranged from 16 to 122 movements per hours of sleep, was not significantly correlated with ESS scores (r = 0.049, n = 18, p > 0.1).
DISCUSSION These results provide evidence that a questionnairebased scale as brief and as simple as the ESS can give valid measurements of sleep propensity in adults. ESS scores significantly distinguished groups of patients who are known from other investigations to have differences in their levels of sleepiness, as measured by the MSLT (2,18). ESS scores were significantly correlated Sleep. Vol. 14. No.6. 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
used to test the differences in ESS scores between primary snorers and the three categories of OSAS. The ScheWe test is conservative and is suitable for groups with unequal numbers of subjects (17). The distribution of sleep latencies ,measured in minutes, was highly skewed positively and was normalized by log., transformation. The relationships between pairs of continuous variables, such as RDI and sleep latency during overnight polysomnography, were tested by Pearson correlation coefficients and linear regression. Statistical significance was accepted at p < 0.05 in two-tailed tests.
543
544
M. W. JOHNS
with sleep latency measured during the day with MSLTs and at night with polysomnography. This is despite any effect of the first night in the laboratory. Others have found a significant positive correlation between the SL at night and during the day in the same subject (19).
ESS scores greater than 16, indicative of a high level of daytime sleepiness, were encountered only in patients with moderate or severe OSAS (RDI > 15), narcolepsy or idiopathic hypersomnia. These disorders are known to be associated with excessive daytime sleepiness as measured by the MSLT (2,18). Newrtheless, high ESS scores, by themselves, are not diagnostic of a particular sleep disorder, any more than is an SL of 5 minutes in an MSLT. ESS scores were correlated with both the RDI and the minimum Sa02 recorded during polysomnography Acknowledgement: Irene Lehel assisted with the adminin patients with OSAS of differing severity. In the past, these measures of the severity of OSAS have been found istration of questionnaires to the control subjects. to be related to the SL in MSLTs in some, but not in 1. American Sleep Disorders Association. The international clasall investigations (18,20). The finding that ESS scores sification of sleep disorders. Rochester, MN, 1990. can distinguish patients who simply snore from those 2. Richardson G, Carskadon M, Flagg W, Van Den Hoed J, Dement W, Mitler M. Excessive daytime sleepiness in man: mulwith even mild OSAS is evidence for the sensitivity of tiple sleep latency measurements in narcoleptic vs. control subthe ESS. The questionnaire should be useful in elucijects. Electroencephalogr Clin NeurophysioI1978;45:621-7. dating the epidemiology of snoring and OSAS, and any 3. Carskadon MA, Dement WC. The multiple sleep latency test: what does it measure? Sleep 1985;5:S67-72. associated cardiovascular or cerebrovascular risks. 4. Zwyghuizen-Doorenbos A, Roehrs T, Schaeffer M, Roth T. TestPrevious investigations of this kind have tended to blur retest reliability of the MSLT. Sleep 1988;11:562-5. the distinction between primary snoring and OSAS 5. Mitler M, Gujavarty KS, Browman CPo Maintenance of wakefulness test: a polysomnographic technique for evaluating treat(21). ment in patients with excessive somnolence. Electroencephalogr In the patients with PLMD, the finding of an almost Clin Neurophysiol1982; 153:658-61. 6. Erman MK, Beckham B, Gardner DA, Roffwarg HP. The modzero correlation between their periodic movement inified assessment of sleepiness test (MAST). Sleep Res 1987; 16: dex and ESS scores suggests that whatever level of 550. daytime sleepiness is associated with PLMD, it is not 7. Pressman MR, Fry JM. Relationship of autonomic nervous system activity to daytime sleepiness and prior sleep. Sleep 1989; 12: related simply to the frequency oflimb movements. It 239-45. may be more closely related to the frequency of those 8. Broughton R, Aguirre M, Dunham W. A comparison of multiple movements producing arousal rather than those that and single sleep latency and cerebral evoked potential (P300) measures in the assessment of excessive daytime sleepiness in do not. This distinction was not made here and further narcolepsy-{:ataplexy. Sleep 1988;11:537-45. investigation is required to clarify this relationship. 9. Johnson LC, Spinweber CL. Gomez SA, Matteson LT. Daytime The low ESS scores of patients with idiopathic or sleepiness, performance, mood, nocturnal sleep: the effect of psychophysiological insomnia are consistent with evbenzodiazepine and caffeine on their relationship. Sleep 1990; 13: 121-35. idence that such patients have a low sleep propensity, 10. Hoddes E, Zarcone V, Smythe H, Phillips R, Dement WC. even when they are able to relax (22). It must not be Quantification of sleepiness: a new approach. Psychophysiology assumed, however, that this is necessarily so for other 1973;10:431-6. 11. Herbert M, Johns MW, Dor€: C. Factor analysis of analogue kinds of insomnia, such as with mood disorders. scales measuring subjective feelings before and after sleep. Br J The relatively wide range ofESS scores in the control Med PsychoI1976;49:373-9. subjects [2-10] is consistent with evidence that some 12. Cook Y, Schmitt F, Berry D, Gilmore R, Phillips B, Lamb D. The effects of nocturnal sleep, sleep disordered breathing and healthy adults, without recognizable sleep disordt~rs, periodic movements of sleep on the objective and subjective remain sleepier than others during the day (23). Such assessment of daytime somnolence in healthy aged adults. Sleep differences persist in MSLTs, even after extending the Res 1988;17:95. hours of nocturnal sleep to overcome possible sleep 13. Johns MW, Gay TJA, Goodyear MDE, Masterton JP. Sleep habits of healthy young adults: use of a sleep questionnaire. Br deprivation (24). The sleep propensity of a subject on J Prev Soc Med 1971;25:236-41. a particular day would be influenced by the quality and 14. Billiard M, Alperovitch A, Perot C, Jammes A. Excessive daytime somnolence in young men: prevalence and contributing duration of prior sleep or of sleep deprivation, the time factors. Sleep 1987; 10:297-305. of day, the presence of various sleep disorders, drug 15. Roth T, Roehrs T, Carskadon M, Dement W. Daytime sleepiness and alertness. In: Kryger MH, Roth T, Dement WC, eds. effects, the level of interest and motivation induced by Sleep, Vol. 14, No.6, 1991
Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
the situation at hand, as well as longer-term physiological differences. The ESS does not distinguish the nature oflong-term physiological or pathological processes that produce a particular level of sleep propensity. Other investigations, including overnight polysomnography, are required for that. The ESS assumes that subjects can remember whether or not and under what circumstances they have dozed off during the day as part of their "usual way of life in recent times". The present results suggest that most patients can give meaningful self reports about this aspect of their behavior and that their ESS scores provide a measurement of their general level of daytime sleepiness, from low to very high levels. This has not been achieved previously by any other published questionnaire.
A SCALE MEASURING DAYTIME SLEEPINESS
16. 17. 18. 19. 20.
Principles and practice of sleep medicine. Philadelphia: Saunders, 1989:14-23. Schmidt-Nowara WW, Wiggins CL, Walsh JK, Bauer C. Prevalence of sleepiness in an adult population. Sleep Res 1989;18: 302. Munro BH, Yisintainer MA, Page EB. Statistical methods for health care research. Philadelphia: J. B. Lippincott Co., 1986. Roehrs T, Zorick F, Wittig R, Conway W, Roth T. Predictors of objective level of daytime sleepiness in patients with sleeprelated breathing disorders. Chest 1989;95:1202-6. Kaplan J, Fredrickson PA, Renaux SR. Nighttime sleep latency as an indicator of daytime sleepiness. Sleep Res 1990; 19:241. Guilleminault C, Partinen M, Quera-Salva MA, Hayes B, De-
21. 22. 23. 24.
545
ment WC, Nino-Murcia G. Determinants of daytime sleepiness in obstructive sleep apnea. Chest 1988;94:32-7. Waller PC, Bhopal RS. Is snoring a cause of vascular disease: an epidemiological review. Lancet 1989;1:143-6. Stepanski E, Zorick F, Roehrs T, Young D, Roth R. Daytime alertness in patients with chronic insomnia compared with asymptomatic control subjects. Sleep 1988; II :54-60. Lavie P, Segal S. Twenty-four-hour structure of sleepiness in morning and evening persons investigated by ultrashort sleepwake cycle. Sleep 1989;12:522-8. Roehrs T, Timms Y, Zwyghuizen-Doorenbos A, Roth T. Sleep extension in sleepy and alert normals. Sleep 1989;12:449-57. Downloaded from https://academic.oup.com/sleep/article-abstract/14/6/540/2742871 by guest on 02 October 2018
Sleep, Vol. 14, No.6, 1991
